Asymptotic-Preserving Discrete Schemes for Non-Equilibrium Radiation Diffusion Problem in Spherical and Cylindrical Symmetrical Geometries

Abstract

We study the asymptotic-preserving fully discrete schemes for nonequilibrium radiation diffusion problem in spherical and cylindrical symmetric geometry. The research is based on two-temperature models with Larsen's flux-limited diffusion operators. Finite volume spatially discrete schemes are developed to circumvent the singularity at the origin and the polar axis and assure local conservation. Asymmetric second order accurate spatial approximation is utilized instead of the traditional first order one for boundary flux-limiters to consummate the schemes with higher order global consistency errors. The harmonic average approach in spherical geometry is analyzed, and its second order accuracy is demonstrated. By formal analysis, we prove these schemes and their corresponding fully discrete schemes with implicitly balanced and linearly implicit time evolutions have first order asymptoticpreserving properties. By designing associated manufactured solutions and reference solutions, we verify the desired performance of the fully discrete schemes with numerical tests, which illustrates quantitatively they are first order asymptotic-preserving and basically second order accurate, hence competent for simulations of both equilibrium and non-equilibrium radiation diffusion problems.

A New Relaxed Splitting Preconditioner for Multidimensional Multi-Group Radiation Diffusion Equations

“…In order to evaluate the nonlinear parts at the previous nonlinear iteration level, Kačanov [29] employed the method of frozen coefficients. An important observation is that because of the complex volatile nonlinear coupling of various physical quantities from numerous interacting spatial and temporal scales, the MGD equations are often discretised by finite volume schemes [14,24,38,45,47,52]. This leads to the series of nonsymmetric but positive definite linear systems, which have to be solved at each time-step and/or nonlinear iterations.…”